At ASELSAN, I develop computer vision algorithms for antenna scan pattern analysis and classification, from dataset preparation to deep learning model evaluation. I also generate realistic synthetic datasets and train YOLO-based segmentation models.
I develop and maintain signal processing and tracking algorithms, including the tracking component of the processing pipeline. I developed a Temporal Convolutional Network for PRI modulation type classification, validated it in MATLAB, and converted it to C++ for system integration.
For my senior design project with Meteksan Defense, I developed deep learning-based super-resolution methods for Synthetic Aperture Radar imagery using PyTorch. I trained Swin2MoSE-based models and combined their outputs with pixel-level weighted fusion to suppress speckle and preserve fine detail.
